TOPICS
Topic 1: Seamless Navigation between DFGs and Petri Nets using Time-Order Maps (Bachelor/Master)
Description: The directly-follows graph (DFG) is the most widely used process graph in the industry, yet it can lead to misleading analytical results if not well understood. This project investigates how visualization can reduce bias and improve traceability by enabling seamless navigation between two distinct process models: the DFG and the Petri net. We build on a similar visual technique, the Time-Order Map, and evaluate the results using benchmark event data. Basic knowledge of process mining or business process management (BPM) is preferable.
Initial References:
[1] C. Rubensson and J. Mendling, “Time-Order Map for Seamless Zooming between Process Models and Process Instances,” in 2025 7th International Conference on Process Mining (ICPM), Oct. 2025, pp. 1–8. URL: https://doi.org/10.1109/ICPM66919.2025.11220736 (alternative: https://drive.google.com/drive/folders/1fQq_pl9wjW7xTxZ5-PzmbF2g84iYvM4M on p.83).
[2] W. M. P. van der Aalst, “A practitioner’s guide to process mining: Limitations of the directly-follows graph,” Procedia Computer Science, vol. 164, pp. 321–328, 2019, URL: https://doi.org/10.1016/j.procs.2019.12.189.
Supervisor: Christoffer Rubensson
Topic 2: Visualizing Multi-Perspective Event Networks in Time-Order Maps (Bachelor/Master)
Description: The Time-Order Map is a visual technique that maps event networks in 2D coordinates and applies semantic zooming, allowing seamless navigation between the case view and the model view of event data. Currently, this visual technique is limited to event networks depicting only a single entity (e.g., an order in an order-to-cash process), not multiple ones (e.g., an order and a customer in an order-to-cash process). We extend the Time-Order Map to visualize multi-perspective event networks (that is, knowledge graphs) by adapting edge semantics and evaluating the results using benchmark event data. Basic knowledge of process mining or graph data is preferable.
Initial References:
[1] C. Rubensson and J. Mendling, “Time-Order Map for Seamless Zooming between Process Models and Process Instances,” in 2025 7th International Conference on Process Mining (ICPM), Oct. 2025, pp. 1–8. URL: https://doi.org/10.1109/ICPM66919.2025.11220736 (alternative URL: https://drive.google.com/drive/folders/1fQq_pl9wjW7xTxZ5-PzmbF2g84iYvM4M on p.83).
[2] S. Esser and D. Fahland, “Multi-Dimensional Event Data in Graph Databases,” Journal on Data Semantics, vol. 10, no. 1, pp. 109–141, Jun. 2021, URL: https://doi.org/10.1007/s13740-021-00122-1.
Supervisor: Christoffer Rubensson
Topic 3: Multi-LLM-Agent Process Simulation (Master)
Description: Process simulation is a central technique in business process management for analyzing what-if scenarios. Traditional business process simulation approaches rely on explicitly specified control-flow logic, which often leads to oversimplified task execution and limited representation of process dynamics. Recent advances in Large Language Models (LLMs) enable a new class of generative systems in which multiple LLM-based AI agents represent different process roles and interact to produce process outcomes. However, it remains largely unexplored how such LLM-based process simulations can be systematically designed and analyzed. The goal of this thesis is to design and implement a novel multi-LLM-agent process simulation that reproduces a small real-world business process and generates event logs suitable for process mining. The thesis will present a novel simulation artifact (e.g., a lightweight framework and prototype) and evaluate it against traditional business process simulation.
Initial References:
[1] Kirchdorfer, L., Blümel, R., Kampik, T., Van der Aa, H., and Stuckenschmidt, H. 2024. “AgentSimulator: An Agent-Based Approach for Data-Driven Business Process Simulation,” in 2024 6th International Conference on Process Mining (ICPM), pp. 97–104. (https://doi.org/10.1109/ICPM63005.2024.10680660).
[2] Gao, C., Lan, X., Li, N., Yuan, Y., Ding, J., Zhou, Z., Xu, F., and Li, Y. 2024. “Large Language Models Empowered Agent-Based Modeling and Simulation: A Survey and Perspectives,” Humanities and Social Sciences Communications (11:1), p. 1259. (https://doi.org/10.1057/s41599-024-03611-3).
[3] Sargent, R. G. 2013. “Verification and Validation of Simulation Models,” Journal of Simulation (7:1), pp. 12–24. (https://doi.org/10.1057/jos.2012.20).
Supervisor: Lennart Ebert
Topic 4: Predictive Process Monitoring (Master)
Description: Predictive Process Monitoring focuses on anticipating future states of running business processes (e.g., remaining time, next activities, or outcome) based on event data. Students who have taken the course on Process Prediction and Machine Learning are especially encouraged to apply with their own research ideas related to the course content.
Interested students are invited to submit a short pitch including:
- the research problem and question they would like to address, and
- references to relevant papers that motivate or relate to the idea.
Initial References:
[1] Di Francescomarino, C., Ghidini, C. (2022). Predictive Process Monitoring. In: van der Aalst, W.M.P., Carmona, J. (eds) Process Mining Handbook. Lecture Notes in Business Information Processing, vol 448. Springer, Cham. https://doi.org/10.1007/978-3-031-08848-3_10
Supervisor: Kate Revoredo
Topic 5: Transfer Learning for Predictive Process Monitoring of Parliamentary Processes (Bachelor/Master)
Description:
Cross-organizational process mining can help to compare similar processes from different organizations with the aim of facilitating benchmarking and mutual learning. While process mining typically focuses on business processes, we have shown that also parliamentary processes can be analyzed from this perspective.
This thesis aims to explore the potential of transfer-learning approaches for predictive process monitoring using data from parliamentary processes. Simply put, transfer learning approaches are machine-learning techniques allowing for different origins of the models training and testing data sets. The core idea is to exploit structural and semantical similarities across use cases - enabling the transfer of knowledge by training models on available data from one use case and applying the model to other use cases.
Initial references:
[1] Hillmann, Paul-Julius, Stephan A. Fahrenkrog-Petersen, and Jan Mendling. "Cross-Organizational Analysis of Parliamentary Processes: A Case Study." 2025 7th International Conference on Process Mining (ICPM). IEEE, 2025.
[2] Liessmann A., Wang W., Weinzierl S., Zilker S., Matzner M., Transfer Learning for Predictive Process Monitoring, Proceedings of the European Conference on Information Systems, AIS, 2024
[3] Weinzierl S., Zilker S. Liessmann A., Käppel M. Wang W., Matzner M., From Source to Target: Leveraging Transfer Learning for Predictive Process Monitoring in Organizations, Business & Information Systems Engineering, 2025
Supervisor: Paul-Julius Hillmann